Exact averages over subsets of 16 trials gave larger mean auditory tail spread contrasts than full trial pools in 140 selected ds004574 participants

This study tested auditory response time spread in OpenNeuro ds004574 version 1.0.0. The main cohort had 92 people with Parkinson's disease and 48 controls. The analysis treated source RT values as seconds. It used correct times from 150 to 1000 ms. The task scheduled 24 auditory novel trials and 192 standard trials per person. Tail spread was Q90 minus Q50. The method calculated exact averages over all uniform subsets of 16 trials. Compared with full trial pools, this method raised the mean auditory minus standard spread by 9.54 ms in Parkinson's disease and 8.09 ms in controls. This change defines a different finite-sample statistic. It is not a known population bias correction. The primary group model adjusted for age and recorded gender. Its group estimate was 0.75 ms, with a 95% person bootstrap interval of -13.52 to 15.54 ms. The group difference remained unresolved when each block supplied four trials. The uniform order-statistic weights are established mathematics. The package adds a tested implementation and a public task application. It preserves source files, hashes, code, trial status, scores, bootstrap outputs, proofs, and reproduction instructions.

quantiles · response time · trial count · Parkinson disease · bootstrap · subset sampling